libx265: Bump version requirement
[libav.git] / libavcodec / libx265.c
CommitLineData
50ea9315
DB
1/*
2 * libx265 encoder
3 *
4 * Copyright (c) 2013-2014 Derek Buitenhuis
5 *
6 * This file is part of Libav.
7 *
8 * Libav is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
12 *
13 * Libav is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 * Lesser General Public License for more details.
17 *
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with Libav; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 */
22
23#include <x265.h>
24
25#include "libavutil/internal.h"
26#include "libavutil/common.h"
27#include "libavutil/opt.h"
28#include "libavutil/pixdesc.h"
29#include "avcodec.h"
30#include "internal.h"
31
8aca00cc
DB
32#if defined(_MSC_VER)
33#define X265_API_IMPORTS 1
34#endif
35
50ea9315
DB
36typedef struct libx265Context {
37 const AVClass *class;
38
39 x265_encoder *encoder;
40 x265_param *params;
41 uint8_t *header;
42 int header_size;
43
44 char *preset;
45 char *tune;
46 char *x265_opts;
47} libx265Context;
48
49static int is_keyframe(NalUnitType naltype)
50{
51 switch (naltype) {
52 case NAL_UNIT_CODED_SLICE_BLA_W_LP:
53 case NAL_UNIT_CODED_SLICE_BLA_W_RADL:
54 case NAL_UNIT_CODED_SLICE_BLA_N_LP:
55 case NAL_UNIT_CODED_SLICE_IDR_W_RADL:
56 case NAL_UNIT_CODED_SLICE_IDR_N_LP:
57 case NAL_UNIT_CODED_SLICE_CRA:
58 return 1;
59 default:
60 return 0;
61 }
62}
63
64static av_cold int libx265_encode_close(AVCodecContext *avctx)
65{
66 libx265Context *ctx = avctx->priv_data;
67
68 av_frame_free(&avctx->coded_frame);
69 av_freep(&ctx->header);
70
71 x265_param_free(ctx->params);
72
73 if (ctx->encoder)
74 x265_encoder_close(ctx->encoder);
75
76 return 0;
77}
78
79static av_cold int libx265_encode_init(AVCodecContext *avctx)
80{
81 libx265Context *ctx = avctx->priv_data;
82 x265_nal *nal;
0f7fa48c 83 int sar_num, sar_den;
50ea9315
DB
84 int nnal;
85 int ret;
86 int i;
87
2142b2ef
DB
88 if (avctx->strict_std_compliance > FF_COMPLIANCE_EXPERIMENTAL &&
89 !av_pix_fmt_desc_get(avctx->pix_fmt)->log2_chroma_w &&
90 !av_pix_fmt_desc_get(avctx->pix_fmt)->log2_chroma_h) {
91 av_log(avctx, AV_LOG_ERROR,
92 "4:4:4 support is not fully defined for HEVC yet. "
93 "Set -strict experimental to encode anyway.\n");
94 return AVERROR(ENOSYS);
95 }
96
50ea9315
DB
97 avctx->coded_frame = av_frame_alloc();
98 if (!avctx->coded_frame) {
99 av_log(avctx, AV_LOG_ERROR, "Could not allocate frame.\n");
100 return AVERROR(ENOMEM);
101 }
102
103 ctx->params = x265_param_alloc();
104 if (!ctx->params) {
105 av_log(avctx, AV_LOG_ERROR, "Could not allocate x265 param structure.\n");
106 return AVERROR(ENOMEM);
107 }
108
50ea9315
DB
109 if (x265_param_default_preset(ctx->params, ctx->preset, ctx->tune) < 0) {
110 av_log(avctx, AV_LOG_ERROR, "Invalid preset or tune.\n");
111 return AVERROR(EINVAL);
112 }
113
114 ctx->params->frameNumThreads = avctx->thread_count;
d00a504b
DB
115 ctx->params->fpsNum = avctx->time_base.den;
116 ctx->params->fpsDenom = avctx->time_base.num * avctx->ticks_per_frame;
50ea9315
DB
117 ctx->params->sourceWidth = avctx->width;
118 ctx->params->sourceHeight = avctx->height;
d00a504b 119
0f7fa48c
DB
120 av_reduce(&sar_num, &sar_den,
121 avctx->sample_aspect_ratio.num,
122 avctx->sample_aspect_ratio.den, 4096);
d67dbfa7
DB
123 ctx->params->vui.bEnableVuiParametersPresentFlag = 1;
124 ctx->params->vui.bEnableAspectRatioIdc = 1;
125 ctx->params->vui.aspectRatioIdc = 255;
126 ctx->params->vui.sarWidth = sar_num;
127 ctx->params->vui.sarHeight = sar_den;
0f7fa48c 128
d00a504b
DB
129 if (x265_max_bit_depth == 8)
130 ctx->params->internalBitDepth = 8;
131 else if (x265_max_bit_depth == 12)
132 ctx->params->internalBitDepth = 10;
50ea9315 133
2142b2ef
DB
134 switch (avctx->pix_fmt) {
135 case AV_PIX_FMT_YUV420P:
136 case AV_PIX_FMT_YUV420P10:
137 ctx->params->internalCsp = X265_CSP_I420;
138 break;
139 case AV_PIX_FMT_YUV444P:
140 case AV_PIX_FMT_YUV444P10:
141 ctx->params->internalCsp = X265_CSP_I444;
142 break;
143 }
144
50ea9315
DB
145 if (avctx->bit_rate > 0) {
146 ctx->params->rc.bitrate = avctx->bit_rate / 1000;
147 ctx->params->rc.rateControlMode = X265_RC_ABR;
148 }
149
150 if (ctx->x265_opts) {
151 AVDictionary *dict = NULL;
152 AVDictionaryEntry *en = NULL;
153
154 if (!av_dict_parse_string(&dict, ctx->x265_opts, "=", ":", 0)) {
155 while ((en = av_dict_get(dict, "", en, AV_DICT_IGNORE_SUFFIX))) {
156 int parse_ret = x265_param_parse(ctx->params, en->key, en->value);
157
158 switch (parse_ret) {
159 case X265_PARAM_BAD_NAME:
160 av_log(avctx, AV_LOG_WARNING,
161 "Unknown option: %s.\n", en->key);
162 break;
163 case X265_PARAM_BAD_VALUE:
164 av_log(avctx, AV_LOG_WARNING,
165 "Invalid value for %s: %s.\n", en->key, en->value);
166 break;
167 default:
168 break;
169 }
170 }
171 av_dict_free(&dict);
172 }
173 }
174
175 ctx->encoder = x265_encoder_open(ctx->params);
176 if (!ctx->encoder) {
177 av_log(avctx, AV_LOG_ERROR, "Cannot open libx265 encoder.\n");
178 libx265_encode_close(avctx);
179 return AVERROR_INVALIDDATA;
180 }
181
182 ret = x265_encoder_headers(ctx->encoder, &nal, &nnal);
183 if (ret < 0) {
184 av_log(avctx, AV_LOG_ERROR, "Cannot encode headers.\n");
185 libx265_encode_close(avctx);
186 return AVERROR_INVALIDDATA;
187 }
188
189 for (i = 0; i < nnal; i++)
190 ctx->header_size += nal[i].sizeBytes;
191
60728e8b 192 ctx->header = av_malloc(ctx->header_size + FF_INPUT_BUFFER_PADDING_SIZE);
50ea9315
DB
193 if (!ctx->header) {
194 av_log(avctx, AV_LOG_ERROR,
195 "Cannot allocate HEVC header of size %d.\n", ctx->header_size);
196 libx265_encode_close(avctx);
197 return AVERROR(ENOMEM);
198 }
199
21a56117 200 memcpy(ctx->header, nal[0].payload, ctx->header_size);
50ea9315 201
60728e8b
MN
202 if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
203 avctx->extradata_size = ctx->header_size;
204 avctx->extradata = ctx->header;
205
206 ctx->header_size = 0;
207 ctx->header = NULL;
208 }
209
50ea9315
DB
210 return 0;
211}
212
213static int libx265_encode_frame(AVCodecContext *avctx, AVPacket *pkt,
214 const AVFrame *pic, int *got_packet)
215{
216 libx265Context *ctx = avctx->priv_data;
217 x265_picture x265pic;
218 x265_picture x265pic_out = { { 0 } };
219 x265_nal *nal;
220 uint8_t *dst;
221 int payload = 0;
222 int nnal;
223 int ret;
224 int i;
225
41836c4e
DB
226 x265_picture_init(ctx->params, &x265pic);
227
50ea9315
DB
228 if (pic) {
229 for (i = 0; i < 3; i++) {
230 x265pic.planes[i] = pic->data[i];
231 x265pic.stride[i] = pic->linesize[i];
232 }
233
d00a504b
DB
234 x265pic.pts = pic->pts;
235 x265pic.bitDepth = av_pix_fmt_desc_get(avctx->pix_fmt)->comp[0].depth_minus1 + 1;
50ea9315
DB
236 }
237
238 ret = x265_encoder_encode(ctx->encoder, &nal, &nnal,
239 pic ? &x265pic : NULL, &x265pic_out);
240 if (ret < 0)
241 return AVERROR_UNKNOWN;
242
243 if (!nnal)
244 return 0;
245
246 for (i = 0; i < nnal; i++)
247 payload += nal[i].sizeBytes;
248
249 payload += ctx->header_size;
250
251 ret = ff_alloc_packet(pkt, payload);
252 if (ret < 0) {
253 av_log(avctx, AV_LOG_ERROR, "Error getting output packet.\n");
254 return ret;
255 }
256 dst = pkt->data;
257
258 if (ctx->header) {
259 memcpy(dst, ctx->header, ctx->header_size);
260 dst += ctx->header_size;
261
262 av_freep(&ctx->header);
263 ctx->header_size = 0;
264 }
265
266 for (i = 0; i < nnal; i++) {
267 memcpy(dst, nal[i].payload, nal[i].sizeBytes);
268 dst += nal[i].sizeBytes;
269
270 if (is_keyframe(nal[i].type))
271 pkt->flags |= AV_PKT_FLAG_KEY;
272 }
273
274 pkt->pts = x265pic_out.pts;
275 pkt->dts = x265pic_out.dts;
276
277 *got_packet = 1;
278 return 0;
279}
280
281static const enum AVPixelFormat x265_csp_eight[] = {
282 AV_PIX_FMT_YUV420P,
2142b2ef 283 AV_PIX_FMT_YUV444P,
50ea9315
DB
284 AV_PIX_FMT_NONE
285};
286
287static const enum AVPixelFormat x265_csp_twelve[] = {
288 AV_PIX_FMT_YUV420P,
2142b2ef 289 AV_PIX_FMT_YUV444P,
50ea9315 290 AV_PIX_FMT_YUV420P10,
2142b2ef 291 AV_PIX_FMT_YUV444P10,
50ea9315
DB
292 AV_PIX_FMT_NONE
293};
294
295static av_cold void libx265_encode_init_csp(AVCodec *codec)
296{
297 if (x265_max_bit_depth == 8)
298 codec->pix_fmts = x265_csp_eight;
299 else if (x265_max_bit_depth == 12)
300 codec->pix_fmts = x265_csp_twelve;
301}
302
303#define OFFSET(x) offsetof(libx265Context, x)
304#define VE AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM
305static const AVOption options[] = {
306 { "preset", "set the x265 preset", OFFSET(preset), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
307 { "tune", "set the x265 tune parameter", OFFSET(tune), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
308 { "x265-params", "set the x265 configuration using a :-separated list of key=value parameters", OFFSET(x265_opts), AV_OPT_TYPE_STRING, { 0 }, 0, 0, VE },
309 { NULL }
310};
311
312static const AVClass class = {
313 .class_name = "libx265",
314 .item_name = av_default_item_name,
315 .option = options,
316 .version = LIBAVUTIL_VERSION_INT,
317};
318
319AVCodec ff_libx265_encoder = {
320 .name = "libx265",
321 .long_name = NULL_IF_CONFIG_SMALL("libx265 H.265 / HEVC"),
322 .type = AVMEDIA_TYPE_VIDEO,
323 .id = AV_CODEC_ID_HEVC,
324 .init = libx265_encode_init,
325 .init_static_data = libx265_encode_init_csp,
326 .encode2 = libx265_encode_frame,
327 .close = libx265_encode_close,
328 .priv_data_size = sizeof(libx265Context),
329 .priv_class = &class,
330 .capabilities = CODEC_CAP_DELAY | CODEC_CAP_AUTO_THREADS,
331};